Computer Vision

As part of the course “Artificial Intelligence 1”, I worked with Arne Matthes and Timo Wiesner on several computer-vision projects.

The assignments began with classical search algorithms and were gradually extended with machine learning. We trained models ranging from Random Forests and Support Vector Machines to deep neural networks on image data collected at our institute.

The goal was a virtual robot capable of self-localization within institute rooms and path planning. The best model achieved an orientation accuracy of 96% and reliably highlighted salient image features used for decision-making.

Technologies: PyTorch, Machine Learning, Data Transformation